Create presenter videos, translate footage, and build visual agents that answer questions. D-ID offers Studio creation, API integration, and self-hosted avatar deployments.
Used this tool? Rate it
Used this tool? Rate it
D-ID creates avatar-driven videos from text, audio, or still images for marketing, training, social media, and internal communications. An AI avatar can deliver a prepared script as an exported video, while a Visual Agent responds to a viewer in a live conversation. Agentic Videos add that question-and-answer interaction to an existing video. These formats serve related communication tasks, but the viewing experience and the resources being consumed differ.
The Studio can animate a library presenter, an uploaded facial image, or a portrait created with its text-to-image generator. A supplied image establishes the visible presenter, while the script or audio supplies what that presenter says. Choosing a stock presenter avoids the separate process of creating a personal video avatar; uploading a face does not by itself establish permission to use that person’s identity.
Visual Agents connect to existing systems to schedule meetings, retrieve information, and automate workflows inside a conversation.
Agentic Videos combine an uploaded video with a chosen video agent and a defined interaction starting point. Viewers can ask questions through voice or chat during playback, with answers connected to the video script and supplemental knowledge.
Agentic Videos capture audience questions so content teams can understand what viewers ask. The company promotes answers grounded in the script and additional knowledge. That describes the intended retrieval context; it does not establish independent proof that every generated response is accurate, complete, or an authorized statement for a business.
Studio video creation starts with Create video, then an avatar, typed text or uploaded audio, a voice selection, and Generate video. The documented workflow is:
Completed videos can be downloaded or shared from the gallery’s three-dot menu; draft videos cannot be downloaded.
Subtitles require Pro or above, are enabled during generation, and export as a separate SRT file rather than being embedded in the video.
Agent grounding can use uploaded sources only, prioritize them while allowing general knowledge, or operate without custom data. These modes change the relationship between uploaded business material and the model’s other knowledge. A source-only configuration and a hybrid configuration are different choices even when the avatar’s face and voice remain identical.
The Agent setup accepts PDF, TXT, and PPTX, with up to five documents of 500,000 characters each; direct input text supports up to 80,000 characters. The setup guidance favors paragraph-style material, such as articles and FAQs, over relying on tables or images.
Agent setup includes a role, optional instructions, personality, welcome message, starter questions, avoided topics, and maximum answer length.
Studio learning use cases include explainer videos, avatar-led tutorials, and onboarding videos. Prepared presenter videos fit explanations that have an approved script. Existing recordings can instead be localized through video translation when the task is to retain the original speaker while changing the spoken language and matching lip movements.
Agentic Video scenarios include training viewers asking for context, prospects asking product-demo questions, and tutorial users requesting troubleshooting help. A training viewer can ask for additional context without leaving the video experience, while a product viewer can ask about a feature as it appears in a demonstration. These are vendor-described interaction patterns, not measured promises of higher retention, conversion, or resolution rates.
Studio marketing use cases include branded campaigns, announcements, and social posts.
Creative Reality Studio is available on desktop and mobile. This describes the available access routes, not complete feature parity across device interfaces. A created Agent can be shared through a direct link or embedded on a website.
Self-hosted Expressive Avatars run in the customer’s Azure, AWS, or Google Cloud account, using GPU rendering capacity and standard streaming-backend compute.
For self-hosted deployments, D-ID says conversations, configurations, and assets stay inside the customer network and nothing is sent to D-ID while an avatar runs. That is the vendor’s description of the self-hosted runtime. It is not a general statement that content uploaded to the hosted web Studio stays solely on the uploader’s device, nor does it erase the policies of any speech or language provider the customer chooses to connect.
Self-hosted session capacity scales with the customer’s infrastructure. The self-hosted annual platform license includes streaming minutes and supports additional on-demand minutes. “No cap” is qualified by the capacity the deployment can actually support. It is not unlimited compute or an identical purchase to a public Studio subscription, and the quote-based infrastructure offering is separate from the Studio prices below.
The November 1, 2026 public Studio capture displayed annual billing. The following are the displayed starting packages, including separately stated monthly-equivalent headlines and annual charges. They are annual-billing labels, not month-to-month subscription quotes; no annual total is calculated by multiplying a rounded headline.
| Plan | Displayed annual-billing price | Included video-related allowance | Branding | Other displayed allocation |
|---|---|---|---|---|
| Trial | $0 for a 14-day trial | 3 minutes across Videos, Agents, Video Translate, and API | Full-screen watermark | 3 Agentic Videos; personal-use license |
| Lite | $4.70/month; $56 billed annually | 10 min./month across the listed products | D-ID watermark | 3 Agentic Videos |
| Pro | $16/month; $191 billed annually | 15 min./month | AI watermark | 1 voice clone; 5 Agentic Videos |
| Advanced | $108/month; $1,293 billed annually | 100 min./month | Custom logo or blank | 3 voice clones; 10 Agentic Videos |
| Enterprise | Sales quote | Unlimited video minutes as listed, with custom Campaign, Agent, and API minutes | Custom logo | Custom Agentic Video count; team collaboration and enterprise support options |
Separately, Product Terms §4.2 says every Animation bears a synthetic mark, with extra watermarks for Guest Users. Removing, hiding or minimizing these marks requires D-ID’s prior written approval. Enterprise customers can request an alternate watermark to replace the D-ID logo, subject to the same approval.
Lite, Pro, and Advanced advertise unlimited videos for the first month, subject to reasonable-use limits and the Fair-Use Policy. Lite, Pro, and Advanced offer higher credit packages; the listed allowances are the displayed starting allocations. The offer applies to first-month videos. The table shows the recurring starting allowances.
Video consumption rounds up to the next 15-second interval; one minute and 10 seconds consumes one minute and 15 seconds, and unused minutes do not roll over.
Assuming the user owns rights to uploaded images, audio, and text, Trial and Lite are personal-use plans, while Pro, Advanced, and Enterprise allow commercial video use. Paid access alone is insufficient: Lite still has a personal-use boundary. A commercial license covers the permitted use of the generated video under the plan, not a license to arbitrary third-party likenesses, recordings, trademarks, or copyrighted scripts.
Subscriptions renew automatically, and the product terms require cancellation 15 days before the end of the billing period to avoid the next charge. The Studio’s cancellation path is Billing → Cancel plan. The product terms describe payments as irrevocable and nonrefundable except where explicitly provided in the agreement. These are published product terms, rather than a claim that a negotiated enterprise contract, app-store purchase, or mandatory legal right has identical wording.
HeyGen offers personal-avatar video creation and video translation with voice cloning and lip-sync.
Synthesia combines video creation, translation, and distribution with quizzes, branching paths, and SCORM publishing for learning workflows. This gives training teams a separate set of course-delivery and interaction features to compare. D-ID’s own mix includes authored avatar clips, conversational Visual Agents, and question-enabled Agentic Videos.
Users retain ownership of submissions, outputs, and animations while granting D-ID a broad worldwide, perpetual, transferable license to submissions, subject to the agreement and Privacy Policy.
The product terms do not guarantee confidentiality of submissions, animations, or outputs, whether or not users share them. The agreement also reserves moderation and removal powers. D-ID’s terms prohibit unauthorized materials, impersonation, pornography, and harmful uses, and restrict software use to people over 18.
D-ID’s biometric policy says it may collect, process and store face geometry, voiceprints, gait and related features for avatar services, identity verification, fraud prevention and moderation. D-ID may use third-party providers for verification or moderation and share the biometric information with them.
The biometric policy says Biometric Information is not used to improve D-ID’s products or train the models powering them. The policy retains biometric information for avatar creation and maintenance or the maximum time allowed by law, whichever comes first, unless the law requires a shorter period. It says the data is then permanently destroyed. The retention language is purpose- and law-based, rather than a single fixed number of days covering every biometric record.
The privacy policy describes commercially reasonable efforts to follow its retention schedule for non-customer personal data and Applicative Data. It describes automatic erasure of Applicative Data uploaded through the API that users have not deleted when a job finishes, through AWS settings within 24 hours by default, within 72 hours for Agents, or within 14 days for Insight Service. These conditional API windows are different from persistent knowledge documents and files deliberately retained for later reuse.
An API persist flag keeps data on encrypted AWS S3 until the user deletes it; Agent knowledge documents are persistent too. Applicative Data uploaded through the web Studio persists until the user erases it in the Studio.
The privacy policy says that after 180 consecutive days without logging in, user data is deleted after another 14 days and an email notice. D-ID may also delete information 14 days after a contract or specific service expires and after sending an email notice. This is separate from deleting an individual persistent item.
While Applicative Data awaits processing or deletion, the policy says it is not accessed for model training; authorized operators may access it for support or operational service tickets. The no-training wording in this section has that Applicative Data storage context. It is not a blanket claim that no customer information is processed for service operation, identity verification, moderation, or other purposes described in the applicable policies.
API users can optionally provide a signed URL for input in their own AWS S3 storage and separately provide one for output. If they supply the output URL, D-ID’s policy says the result is not stored on its systems.
Cloning another person’s voice requires their explicit permission. They must read and agree to the consent statement during cloning.
After subscription cancellation, the plan stays active until the billing period ends and the account is then downgraded.
